►
Description
SIG Cluster Lifecycle - Cluster API Provider AWS Office Hours - 20201214
A
Okay,
this
is
the
cluster
api
provider
aws
meeting
of
december.
The
14th
2020,
we're
abiding
by
the
cncf
code
of
conduct
broadly
means
be
excellent
to
each
other,
just
throw
the
link,
I
think,
just
for
peter
joined
a
bit
more.
A
few
people
are
joining
so
just
for
the
link
again
new
doc,
so
there's
a
new
link
for
the
doc,
so
click
through
to
that
rather
than
whatever's
in
your
calendar,
and
why
we
will
update
the
calendar
soon.
A
As
just
let
you
know,
we've
had
a
g
suite
migration
at
vmware,
which
means
the
docs
are
basically
being
copied
to
a
new
domain.
We
lost
access
to
the
old
ones,
so
we
need
to
update
everywhere
that
these
old
docs
were
being
used.
So
there's
prs
going
into
kate's
community
github
group,
which
should
then
update
the
calendar
or,
if
not
tim,
said
claire
will
update
it.
A
B
A
Okay,
so
I
think
the
context
of
this
is
we
run
alpha.
Paul
is
not
going
to
be
out
for
quite
a
while,
because
we're
gonna
there's
enough
work
to
do
meanwhile
before,
which
means
it's
not
it's
not
something
we're
going
to
have
ready
in
january
or
whatever.
So,
there's
a
timing
question
here
and
the
decision
it
is
we
we
discussed
this
last
week
and
we
were
kind
of
saying
we
went
alpha.
A
A
Then
main
branches
on
is
on
the
latest.
We
can
test
against
the
tip
of
tree
of
cluster
api.
The
drawback
is:
if,
if
we
want
to
back
port
any
features,
then
we
have
to
backport.
B
And
I
wanna
add
this
here:
since
controller
runtime
versions
will
be
different
between
previous
I
mean
non-master
branch
and
master
branch.
The
back
parts
will
be
a
bit
challenging
e
and
and
the
forward
part.
If
yeah
I
mean
it
goes
worthless.
B
So
yeah,
with
the
controller
runtime
changes
I
mean
yeah
context
are
one
thing:
another
thing
is
enqueuing
functions
are
different
other
than
that
some
tests,
even
tests,
have
some
differences.
A
I
guess
it
makes,
makes
life
easier
in
terms
of
contributing
features,
but
then
we
also,
but
we
need
to
forward
for
everything
with
the
same
with
broadly
the
same
kind
of
challenges
with
the
controller
one.
A
A
C
C
C
B
So
actually,
there
are
a
lot
of
changes
going
on
in
campus
site.
So
if
we
open
the
master
to
v1
alpha
4,
we
we
can
continuously
add
those
changes.
So
we
do
not
miss
anything
going
on
in
the
kept
side
I
mean.
If
we
stay
on
v1
alpha
3,
we
will
actually
do
the
work
once
to
port
everything
at
once.
I
think
it
feels
like
easier
too,
but
we
may
need
to
go
back
to
cappy
changes
and
apply
those
at
the
end
when
we
open
v1
alpha
4.
A
C
A
Cappy
load
balancer
proposal,
which
ties
into
the
kappa
we
want
alpha
for
design,
but
we
haven't
even
designed
those
yet
so
yeah
that
we
they're
we're
not
gaining
much
in
terms
of
the
api
design.
Yet
for
we
went
out
before
it's
still
up
in
air.
A
B
I
mean
we
can
actually
wait
for
maybe
a
couple
of
months
because,
for
example,
even
the
small
version
changes
in
controller
runtime
required
a
change
on
our
site,
so
cape
right
now
is
actively
upping
their
controller
runtime
to
every
alpha
dot
to
alpha.5
alpha.8
release
for
controller
runtime,
and
this
requires
consistent
change
on
our
side.
B
A
Okay,
I
mean
yeah,
I
I
feel
bad
like
you've
done
all
that
work
and
then
we're
saying
we're
not
right
have
it
so
do
you
want
to
do
the
branch
at
all
or
like
just
or
close
the
pr
and
oh.
B
You
mean
like
a
separate
branch,
not
the
master
sure.
If
it,
it
will
help
us
yeah.
A
A
We
want
alpha
3
for
main
branch
and
have
a
go
and
see
see
where
we
get
to,
and
we've
got
some
work
to
do
in
the
cluster
api
repo
anyway,
because
we
don't
have
that
alpha,
we
don't
have
any
way
of
getting
a
release
for
end-to-end
testing
right
for
cappy
yet,
and
we
discussed
an
idea
in
last
week's
meeting
to
publish
like
a
nightly
build.
So
why
don't
we?
A
Why
don't
we
do
the
release?
04
branch,
maybe
do
the
test
and
submit
a
what,
if,
if
they
do
like
that,
alpha
zero
release
for
cappy
set
up
the
testing
for
job?
A
C
A
D
A
C
A
Never
been
never
had
any
more
success
than
that,
which
is
why
you
tend
to
define
as
much
of
it
as
possible
inside
the
cap
and
replay.
So
what
goes
in
testing
is
fairly
minimal
in
terms
of
setup.
I'm
sure
we're
going
to
have
to
do
things
fiddle
with
like
iam
and
stuff
like
that,
but
we
can
sort
that.
A
Yes,
there's
a
bunch
of
them
that
need
updating,
there's
like
a
weird
maintainers
one
as
well.
So
that's
the
one
that
actually
lets
you
do.
The
github
releases
so
yeah
remind
me
tomorrow
about
it
I'll
go
through
it
find
them,
because
I
can't
even
remember
where
all
the
ones
are,
but
there's
one
thing:
hates
io
that
need
update,
that
you
have
permissions
to
do
image
promotion
as
well,
and
things
like
that.
C
Scaling
I
have
permissions
so
without
winning
the
eks.
Today
I
knew
just
for
not
today.
Yesterday
I
had
a
problem
with
the
emergency
problems.
C
C
Has
changes
and
changes
to
the
plus
aws
bootstrap
stuff.
A
Yeah,
I
think
that
should
be
fine
if
it
isn't
I'll.
Let
you
know
I'll.
A
B
A
A
A
Stop
say
anything
else:
anyone
to
bring
up
we've
got
yeah,
pretty
sure.
If
not,
I
can
call
it
quits.